The safety margin of variable withdrawals...

[Cb]

Most of the long-time posters here are probably familiar with various studies of variable withdrawal rates in retirement. For any of you who haven't seen those studies here's an Excel file that demonstrates the extra margin of safety that allowing some portion of your annual withdrawal to float based on portfolio performance:

http://gnobility.com/Finance/SWR_using_Variable_Withdrawals.xls

Set up an old-school 30 or 40 year retirement using a basic S&P 500 - Bond split, then use the little arrow keys at cell E19 to dial up ever larger "percentage of portfolio" draws, and watch the impact that it has on the Total SWR.

Allowing 0.50% to 1.00% of your annual draw to vary based on portfolio size provides a fair amount of cushion, and wouldn't be all that painful in practice - you just defer auto replacements, more lavish vacations, and remodeling projects until your portfolio recovers.

Cb
PS: this app's results are within a few tenths of the SWR's provided by FIREcalc. You can model the same thing in FIREcalc using SG's trick of cranking the FIREcalc's Portfolio Expense Ratio input block up (you must add to whatever ER your portfolio will have...i./e. ~0.22% ER + 0.75% Percent of Port)

 

[MasterBlaster]

It may be enlightening and interesting to see some sort of visualization of survival rates as a function of various depletion models. I would caution against deciding on one scheme over another based on a few percentage points difference in their survivor rates.

The variation of the underlying data is such that finely-tuned conclusions of various depletion models is somewhat dangerous.

If you can get to within a few percent of where you would like to be (in terms of survival rates) then consider any depletion scheme with similar results to be successful.

 

[samclem]

There have been some good threads in the past that discussed the likely real-life impact of a "near miss." When you run FIRECalc or any other tool and get a 95% survival rate, it would be useful to also see just how frequently the time series brought the balance to very low levels. As most of us contemplate recent losses, it might be useful to imagine how it would feel if your portfolio were down 75% and the market was still headed down. As long as it climbs up before hitting a zero balance, this would count as a "success," but it sure wouldn't be a nice ride.

 

[Gardnr]

There have been some good threads in the past that discussed the likely real-life impact of a "near miss." When you run FIRECalc or any other tool and get a 95% survival rate, it would be useful to also see just how frequently the time series brought the balance to very low levels. As most of us contemplate recent losses, it might be useful to imagine how it would feel if your portfolio were down 75% and the market was still headed down. As long as it climbs up before hitting a zero balance, this would count as a "success," but it sure wouldn't be a nice ride.

Exactly. I just wrote something similar in another thread. I think people often (usually?) don't really look at those worst case scenarios and think about what they mean, figuring things could never get that bad again. Oops, that may not be true.

Or maybe they do and figured they'd go back to work if it got that bad; and now they are. Can't blame 'em. And it's one perfectly good way to deal with it.

 

[youbet]

As long as it climbs up before hitting a zero balance, this would count as a "success," but it sure wouldn't be a nice ride.

Yes indeed. This is an aspect of Firecalc output that it took me a while to catch on to. I usually use 100% survival rates when doing Firecalc runs, but even then looking at the outputted spreadsheet I see "close calls."

It must be human nature to assume that your actual retirement results will be at least "average." That's what I used to do. But after many scans of Firecalc output, I've finally come to grips with the fact that a retirement scenario may test as 100% survivable in Firecalc and still be perfectly capable of scaring the begeesits outa ya as you go through retirement.

That's kinda happening to me right now..... I'm 2.3 yrs into RE and my net worth has decided to dive 20 - 25% depending on when I look at it....

Cb

- thanks for the link. Interesting.

 

[donheff]

How do you read the calculator? As I added points to the "variable" component that I understood to show how much less you would take in bad years my SWR dropped and the % taken in good starting years dropped. Am I reading this thing improperly?

 

[Cb]

How do you read the calculator? As I added points to the "variable" component that I understood to show how much less you would take in bad years my SWR dropped and the % taken in good starting years dropped. Am I reading this thing improperly?

Have a look at this thread:

Retire Early Home Page Discussion Board :: View topic - SWR for basic vs. discretionary spending

Cell D19 shows the portion of your annual draw that you're willing to allow to float around based on market performance

Cell D21 shows the 'conventional' FIREcalc-type unwavering initial year 1 draw adjusted for inflation

Cell D23 shows the sum of those two components of your draws

You should focus on the "Total SWR" number in cell D23...the sum of the conventional FIREcalc fixed draw adjusted for inflation, PLUS the percentage of remaining portfolio.

The more you're willing to allow your annual draws to vary according the to whims of the market, the more you can draw from the outset. I think people do that naturally...when the market is plummeting we're less inclined to plan a trip to Hawaii, or buy a new car rather than repair the old one.

Cb

 

[donheff]

Thanks Cb, I see it now. The graph on the right was confusing me.

 

[Independent]

Thanks for the link. I think the idea of nearly 100% certainty on a "basic" withdrawal amount, plus a variable "nice to have" withdrawal makes a lot of sense. Many of us are in the situation where SS in much of our "basic" need, so the fixed portion can be pretty low.

One note with this particular spreadsheet is that the variable portion can be large enough that in "average" scenarios you have a planned reduction in spending (because the variable withdrawal is a percent of the current fund balance). This may be fine - many people expect their spending to go down with age - but you need to think about the issue.

 

[haha]

There have been some good threads in the past that discussed the likely real-life impact of a "near miss." When you run FIRECalc or any other tool and get a 95% survival rate, it would be useful to also see just how frequently the time series brought the balance to very low levels. As most of us contemplate recent losses, it might be useful to imagine how it would feel if your portfolio were down 75% and the market was still headed down. As long as it climbs up before hitting a zero balance, this would count as a "success," but it sure wouldn't be a nice ride.

Nor would most mortals stick around to complete the ride. Mr. Bogle or his successor would be on TV counseling us to bail if "you absolutely cannot afford to lose another penny." Well, I would say that if the surrent dust-up continues on down for another 2000 points or so, many of us without pensions will be in just that spot.

I think unless she has other resources, not many who retired without a very large surplus will be able to maintain during even a 50% cut.

Of couse if she has plenty of cash, and a conservative fixed income allocation of at least 50%, s/he would be OK even with a >50% equity drop.

For me, it comes back to income. Some here posit that dividends fall apart under stress, and have even stated that S&P dividends have fallen by 1/3 this year. Not much can be said about that other than do your own homework and get the facts straight.

A porfolio of SPIAs might offer fairly robust help too.

My anchor to windward is portfolio income, but I concede that there are valid criticisms of this approach. I believe though that there are several factors in its favor- not the least of which is that is has been practiced for hundreds of years.

Ha

 

[tipster]

It may be enlightening and interesting to see some sort of visualization of survival rates as a function of various depletion models. I would caution against deciding on one scheme over another based on a few percentage points difference in their survivor rates.

There are a lot of MC simulators that model variable, portfolio-performance-based, withdrawal strategies.

What I think most people would do -- and indeed, what it looks like people on this forum are doing -- is periodically re-simulate and re-project the retirement expenditures their portfolio can sustain, and adjust their spending accordingly.

I think a sensible withdrawal strategy is to withdraw an amount equal to the greater of (1) an absolute minimum amount (say, $40K/year) and (2) the stream of income payments that your portfolio, if treated like an inflation-adjusted annuity, would provide if it earned your expected rate of return over a long time period (e.g., until you would be 95-100 years old).

I would use the PMT formula in Excel (with NPER = targeted remaining portfolio life) to make that calculation, repeat that calculation every year, and adjust my withdrawal accordingly. This way I would take advantage of good times, and tighten accordingly in bad times.

In fact, that's pretty close to how I simulate planned withdrawals in my own MC simulations.